© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.

org (ISSN-2349-5162)

RECIPE RECOMMENDATION SYSTEM BASED

ON FOOD INGREDIENTS
1
Rajesh D.Potdar, 2S. T. Patil
1
M.Tech. Student ,2 Professor,
CSE Department,
Vishwakrma Institute of Technology, Pune, India.

Abstract: Making decision about what and where to eat is a major problem in our everyday lives due to a wide variety
of ingredients, culinary styles, ethnicities, cultures, and personal tastes. Choosing right dish at the right time seems to be
a very difficult task.This study explains recipe recommendation system based on food ingredients using Knn
algorithm.There are many recipes or food recommendation system but each have their drawbacks. This system uses
users BMI (Body Mass Index) for recommending recipes. Also the food is categorized into three time categories and
two types i.e. veg and non-veg.It will be recommended by using a recommendation algorithm.

Keywords: KNN (K Nearest Neighbors), Recommendation system, Recipe, BMI (Body Mass Index),

1. Introduction
A recommendation system is an information filter system when the information is overloaded by narrowing down the
amount of information to only those interesting content. The recommender system is useful in filtering wide range of
information of products, services, or media content. However, to filter the right information to the customer is a
challenging issue. Normal people usually have three meals a day. For each meal, there are two important questions to
answer which are “what kind of food to eat?” and “Where can we find such meal?”. Although food is essential to our
life, answering the same questions repeatedly is way too boring. It would be helpful to have someone picking our
favorite food or new dish to try out for all our meal in the future. This helper could be developed as a food
recommendation system. We propose web based application which constructs recommendation system by adding more
heterogeneous information of recipes like cuisines, dietary assessment, etc. Recipe recommendation based on food
ingredients is made for users who want healthy recipes for their better health and diet .The recipes are recommended
according to the users’ BMI (Body Mass Index).It will be helpful for all type of users who don’t know about contents of
recipes. There are many businesses deployed the recommendation systems for providing specific content to the
corresponding users such as Amazon [2], YouTube [3], Netflix [4]. Additionally, recommendation system can also be
applied to other fields such as food, news, music, and so on. Therefore, our recommendation will filter food information
to provide the users what to eat for the next meal.

1.1 Problem Statement

To design and implement a system which will recommend recipes by calculating BMI based on food ingredients.

2. Related Work

Currently, there are traditional information retrieval techniques of recommendation [1] which are Content-based
filtering, Collaborative filtering, and Hybrid filtering which can be used for food recommendation. J. Freyne [5]
analyzes performance of those recommendation techniques applied to the food recommendation by comparing the
results of traditional techniques with different inputs. For example, Content-based filtering that utilized user rating,
Collaborative filtering that used similar profile of current user and all users, and Hybrid filtering that combining two
inputs of previous techniques. From their experimental results, the problem occurred when the recipes were broken
down to ingredients. A positive score was assigned to an ingredient with rating, but a negative score was assigned to a
recipe without rating which means dislike in steads of no information available. In 2015, a food recommendation
application was studied using Ingredient tag and Latent factor [7] as inputs by creating a set of ingredients for user to tag
JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 678
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

with a label of like or dislike. This process required many steps of user feedbacks which take a lot of time to tag every
single ingredient. In this paper we have proposed a simple recipe recommendation system based on food ingredients
using Knn (K nearest neighbor).We used users BMI for to recommend their recipes.

3. Proposed Work and Design

The Proposed method consists of three steps that are consisting of input from user; analysis from web application
database. The first step is finding data for dataset. We have used Knn (K nearest Neighbor) algorithm to design this
approach. The users have to add weight and height to calculate their BMI (Body Mass Index).We are Using machine
learning technology to give the output.

3.1 Machine learning

Machine learning is an application of artificial intelligence (AI) that provides systems the ability to
automatically learn and improve from experience without being explicitly programmed. Machine learning focuses on
the development of computer programs that can access data and use it learn for themselves. The process of learning
begins with observations or data, such as examples, direct experience, or instruction, in order to look for patterns in data
and make better decisions in the future based on the examples that we provide. The primary aim is to allow the
computers learn automatically without human intervention or assistance and adjust actions accordingly.

3.2 Flow Diagram of recipe recommendation system based on food ingredients system.

Figure 3.1 Flow Diagram

The diagram shows the flow of the system.

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 679
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

4. Research Methodology

The research methodology includes algorithms used, equations used.

Knn (K nearest Neighbors)

K nearest neighbors is a simple algorithm that stores all available cases and classifies new cases based on a similarity
measure (e.g., distance functions). KNN has been used in statistical estimation and pattern recognition already in the
beginning of 1970’s as a non-parametric technique.

4.1 Knn Algorithm Pseudo code:

1. Calculate “d(x, xi)” i =1, 2… n; where d denotes the Euclidean distance between the points.
2. Arrange the calculated n Euclidean distances in non-decreasing order.
3. Let k be a +ve integer, take the first k distances from this sorted list.
4. Find those k-points corresponding to these k-distances.
5. Let ki denotes the number of points belonging to the ith class among k points i.e. k ≥ 0
6. If ki >kj ∀ i ≠ j then put x in class i.

Figure 4.1 Knn algorithm Example with k values

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 680
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

4.2 Distance Functions

I) Euclidean Distance

√∑𝑛𝑖=1(𝑋𝑖 − 𝑌𝑖)2 …………………………………………(Equation 4.1)

In mathematics, the Euclidean distance or Euclidean metric is the "ordinary" straight-line distance between two points in
Euclidean space. With this distance, Euclidean space becomes a metric space.we have used euclidean distance to
calculate the distance between recipes.

We have differentiated the dataset into three categories i.e. breakfast, lunch and dinner. We have set a time for each type
of food .For breakfast we have set probability limit 0.4, lunch to 0.6, and 0.8 for dinner.

Recommendation Systems algorithm

A recommender system is an Information Filtering (IF) system that provides personal preference guide based on the
user profile and preferences. We have used collaborative filtering algorithm for recommendation with Knn.

5. Results

We have calculated the probability for eating a recipe at the time that is for breakfast, lunch dinner.For example is given
in table.

Type Probability

Breakfast 0.545689

Lunch 0.688955

Dinner 0.826552

Table 5.1 Probability of Food Type

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 681
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

In above screenshot the BMI is calculated after taking inputs of height and weight. After that if you select food
preferences then the recipes list will be generated.

Recipes list is generated then you have to select the recipe for its procedure and contents.

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 682
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

The procedure is shown in the above snapshot. Then after clicking on results the following results will be generated..

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 683
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

Comparison With other algorithms

Algorithm Success rate

Decision Tree classifier 78.86

Naïve Bayes classifier 79.56

KNN classifier 85.74

Table 5.2 Comparison of algorithms with success rate

6. Conclusion

Using Knn algorithm and recommendation algorithm for recipe recommendation we can develop a recipe
recommendation system for healthy living. The recipes can be provided on users health data such as weight,
BMI, etc. The overall accuracy of the system using KNN is 85.74% whereas it is lesser in the other algorithms
such as decision tree classifier which was less than 80%.

References

[1] Yuka SHIDOCHI，Ichiro IDE，Tomokazu TAKAHASHI and Hiroshi MURASE，Finding

Replaceable Materials by Cooking Recipe Mining， IEICE Transactions, Vol. J94-A No.7, 2011,
pp.532-535.

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 684
© 2019 JETIR June 2019, Volume 6, Issue 6 www.jetir.org (ISSN-2349-5162)

[2] Adomavicius G, Tuzhilin A., Towards the next generation of recommender systems: A survey of the
state-of-the-art and possible extensions., IEEE Trans. on Knowledge and Data Engineering,
2005,17(6):734 749. [doi: 10.1109/TKDE.2005.99]

[3] Ricci F., Mobile recommender systems. ,Int’l Journal of Information Technology and Tourism,
2011,12(3):205-231.

[4] Sarwar,Badrul; Karypis,George; Konstan,Joseph; Riedl,John (2001). Item-based collaborative filtering

recommendation algorithms [J]. Proceedings of the 10th international conference on the World Wide
Web (ACM): 285–295.

[5] M. A., Ghazanfar and A. Prugel-Bennett, “A scalable, accurate hybrid recommender system”.
Proceedings of the 2010 Third international conference on Knowledge Discovery and Data Mining,
pp. 94-98, 2010.

JETIR1906V98 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org 685